Product-limit estimators of the survival function with left or right censored data
|
|
- Peter Dorsey
- 5 years ago
- Views:
Transcription
1 Product-limit estimators of the survival function with left or right censored data 1 CREST-ENSAI Campus de Ker-Lann Rue Blaise Pascal - BP Bruz cedex, France ( patilea@ensai.fr) 2 Institut de Statistique Université Catholique de Louvain 20, voie du Roman Pays 1348 Louvain-la-Neuve, Belgique ( rolin@stat.ucl.ac.be) Valentin Patilea 1 and Jean-Marie Rolin 2 Abstract. The problem of estimating the distribution of a lifetime when data may be left or right censored is considered. Two models are introduced and the corresponding product-limit estimators are derived. Strong uniform convergence and asymptotic normality are proved for the product-limit estimators on the whole range of the observations. A bootstrap procedure that can be applied to confidence intervals construction is proposed. Keywords: bootstrap, delta-method, martingales, left and right censoring, strong convergence, weak convergence. 1 Introduction A great deal of recent attention in survival analysis has focused on estimating the survivor distributions in the presence of various and complex censoring mechanisms. The goal of this paper is to analyze simple models for lifetime data that may be left or right censored. Typically, a lifetime T is left or right censored if, instead of observing T we observe a finite nonnegative random variable Y, and a discrete random variable with values 0, 1 or 2. By definition, when A = 0, Y = T, when A = 1, Y < T and, when A = 2, Y > T. Models for left or right censored data were proposed by [Turnbull, 1974], [Sampath and Chandra, 1990] and [Huang, 1999]. See also [Gu and Zhang, 1993], [Kim, 1994]. Assume that the sample consists of n independent copies of (Y, A) and let F T be the distribution of the lifetime of interest T. Using the plug-in (or substitution) principle, the nonparametric estimation of F T is straight as soon as F T can be expressed as an explicit function of the distribution of (Y, A). The existence of such a function requires a precise description of the censoring mechanism that is generally achieved by introducing latent variables and by making assumptions on their distributions. In this paper,
2 980 Patilea and Rolin two latent models allowing for explicit inversion formula, that is closed-form function relating F T to the distribution of (Y, A), are proposed. In some sense, our first latent model lies between the classical rightcensorship model and the current-status data model. It may be applied to the following framework. Consider a study where T the age at onset for a disease is analyzed. The individuals are examined only one time and they belong to one of the following categories: (i) evidence of the disease is present and the age at onset is known (from medical records, interviews with the patient or family members,...); (ii) the disease is diagnosed but the age at onset is unknown or the accuracy of the information about this is questionable; and (iii) the disease is not diagnosed at the examination time. Let C denote the age of the individual at the examination time. In the first case the exact failure time T (age at onset) is observed, that is Y = T. In case (ii) the failure time T is left-censored by C and thus Y = C, A = 2. Finally, the onset time T is right-censored by C for the individuals who have not yet developed the disease; in this case Y = C, A = 1. If no observation as in (ii) occurs, we are in the classical right-censorship framework, while if no uncensored observation is recorded we have current-status data. Our first latent model can be applied, for instance, with the data sets analyzed by [Turnbull and Weiss, 1978]. The second latent model proposed is closely related to the first one. It lies between the left-censorship model and the current-status data model. Consider the example of a reliability experiment where the failure time of a type of device is analyzed. A sample of devices is considered and a single inspection for each device in the sample is undertaken. Some of them already failed without knowing when (left censored observations). To increase the precision of the estimates, a proportion of the devices still working is selected randomly and followed until failure (uncensored observations). For the remaining working devices the failure time is right censored by the inspection time. Let us point out that, without any model assumption, given a distribution for the observed variables (Y, A) with Y 0 and A {0, 1, 2}, we can always apply our two inversion formulae. In this way we build two pseudotrue distribution functions of the lifetime of interest which are functionals of the observed distribution. If the experiment under observation is compatible with the hypothesis of one of our latent models, the true F T can be exactly recovered from the observed distribution. Otherwise, we can only approximate the true lifetime distribution. The paper is organized as follows. Section 2 introduces our two latent models through the equations relating the distribution of the observations to those of the latent variables. Solving these equations for F T we deduce the inversion formulae. The product-limit estimators are obtained by applying the inversion formulae to the empirical distribution. Section 2 is ended with some remarks and comments on related models. Section 3 contains the
3 PL estimators with left or right censored data 981 asymptotic results for the first latent model (similar arguments apply for the second model). We prove strong uniform convergence for the product-limit estimator on the whole range of the observations. Our proof extends and simplifies the results of [Stute and Wang, 1994] and [Gill, 1994] provided in the case of the Kaplan-Meier estimator. Next, the asymptotic normality of our product-limit estimator is obtained. The variance of the limit Gaussian process being complicated, a bootstrap procedure for which the asymptotic validity is a direct consequence of the delta-method is proposed. 2 The latent models 2.1 Model 1 The survival time of interest is T (e.g., the age at onset). Let C be a censoring time (e.g., the age of the individual at the examination time) and be a Bernoulli random variable. Assume that the latent variables T, C and are independent. The observations are independent copies of the variables (Y, A), with Y 0 and A {0, 1, 2}. These variables are defined as Y = min(t, C) + (1 )max(c T, 0) = C + min(t C, 0) and A = 2(1 )1 {T C} +1 {C<T }, where 1 A denotes the indicator function of the set A. With this censoring mechanism the lifetime T is observed, right censored or left censored. In view of the definitions of Y and A, note that if is constant and equal to one (resp. zero), we obtain right censored (resp. current status ) data. Let F T and F C denote the distributions of T and C, respectively. Let p = P ( = 1). Define the observed subdistributions of Y as H k (B) = P (Y B, A = k), k = 0, 1, 2, (1) for any B Borel subset of [0, ]. As usually in survival analysis, the censoring mechanism defines a map Φ between the distributions of the latent variables and the observed distributions. For the censoring mechanism we consider, the relationship (H 0, H 1, H 2 ) = Φ(F T, F C, p) between the subdistributions of Y and the distributions of the latent variables T, C and is the following: H 0 (dt) = p F C ([t, ]) F T (dt) H 1 (dt) = F T ((t, ])F C (dt). (2) H 2 (dt) = (1 p)f T ([0, t])f C (dt) Remark that when p = 1 (resp. p = 0) the equations (2) boil down to the equations of the classical independent right-censoring (resp. current status) model. By plug-in applied with the empirical distribution, the nonparametric estimation of the distribution of T is straight as soon as the map Φ is invertible
4 982 Patilea and Rolin and F T can be written as an explicit function of the observed subdistributions H k, k = 0, 1, 2. The model considered allows us an explicit inversion formula for F T. In order to derive this inversion formula, integrate the first and the second equation in (2) on [t, ] and deduce For t = 0, it follows that H 0 ([t, ]) + ph 1 ([t, ]) = pf T ([t, ])F C ([t, ]). (3) p = H 0([0, ]) 1 H 1 ([0, ]) = H 0 ([0, ]) H 0 ([0, ]) + H 2 ([0, ]). (4) Recall that the hazard measure associated to a distribution F is Λ(dt) = F(dt)/F([t, ]). In our case, use (2)-(3) to deduce that the hazard function corresponding to F T can be written as Λ T (dt) = Finally, the distribution F T can be expressed as H 0 (dt) H 0 ([t, ]) + ph 1 ([t, ]). (5) F T ((t, ]) = [0,t](1 Λ T (ds)), (6) where [0,t] is the product-integral on [0, t]. Note that there is no explicit formula for F T if p = 0 in equations (2), that is with current status data. Given the explicit relationship between the distribution of T and the observed subdistributions, to obtain the product-limit estimator of F T, we simply replace H k, k = 0, 1, 2 by their empirical counterparts. Let F T denote the product-limit estimator of F T. 2.2 Model 2 As in Model 1, assume that T, C and are independent. The observations are independent copies of the variables (Y, A), with Y 0 and A {0, 1, 2} where Y = T, A = 0 if 0 C T and = 1; Y = C, A = 1 if 0 C T and = 0; (7) Y = C, A = 2 if 0 T < C. The equations of this model are H 0 (dt) = p F C ([0, t])f T (dt) H 1 (dt) = (1 p)f T ([t, ])F C (dt). (8) H 2 (dt) = F T ([0, t))f C (dt) Remark that when p = 1 (resp. p = 0) the equations (8) boil down to the equations of the classical independent left-censoring (resp. current status)
5 PL estimators with left or right censored data 983 model. This model also allows for an explicit inversion formula for F T. By integration in the first and the third equation in (8), H 0 ([0, t])+ph 2 ([0, t]) = pf T ([0, t])f C ([0, t]). Deduce p = H 0([0, ]) 1 H 2 ([0, ]). Recall that given a distribution F, the associated reverse hazard measure is M(dt) = F(dt)/F([0, t]). By equations (8) deduce that the reverse hazard function M T associated to F T can be written as M T (dt) = H 0 (dt) H 0 ([0, t]) + ph 2 ([0, t]). Finally, the distribution F T can be expressed as F T ([0, t]) = (1 M T (ds)). (t, ] Applying the inversion formula with the empirical subdistributions, we get the product-limit estimator of F T in Model 2. Note that if T = h(t) and C = h(c), with h 0 a decreasing transformation, then T, C and are the variables of Model 1 applied to the left or right censored lifetime h(y ). In other words, Model 2 is equivalent to Model 1, up to a time reversal transformation. 2.3 Related models [Huang, 1999] introduced a model for the so-called partly interval-censored data, Case 1; see also [Kim, 1994]. In such data, for some subjects, the exact failure time of interest T is observed. For the remaining subjects, only the information on their current status at the examination time is available. [Huang, 1999] considered the nonparametric maximum likelihood estimator (NPMLE) of F T. Unfortunately, NPMLE does not have an explicit form and therefore Huang needs strong assumptions for deriving its asymptotic properties and a numerical algorithm for the applications. Let us point out that, on contrary to our Model 1 (resp. Model 2), in Huang s model one may observe exact failure times even if failure occurs after (resp. before) the examination time. Moreover, in Huang s model one may still obtain a n consistent estimator of the distribution FT if one simply considers the empirical distribution of the uncensored lifetimes. This is no longer true in our models. Perhaps, the most popular model for left or right-censored data is the one introduced by [Turnbull, 1974]; see also [Gu and Zhang, 1993]. In Turnbull s model there are three latent lifetimes L (left-censoring), T (lifetime of interest) and R (right-censoring) with L R. The observed variables
6 984 Patilea and Rolin are Y = max(l, min(t, R)) = min(max(l, T), R) and A defines as follows: A = 0 if L < T R; A = 1 if R < T; and A = 2 if T L. The equations of this model are H 0 (dt) = {F R ([t, ]) F L ([t, ])} F T (dt) H 1 (dt) = F T ((t, ]) F R (dt), H 2 (dt) = F T ([0, t]) F L (dt) where H k, k = 0, 1, 2 are defined as in (1) and F T, F L and F R are the distributions of T, L and R, respectively. The NPMLE of the distribution of the failure time T is not explicit but it can be computed, for instance, by iterations based on the so-called self-consistency equation. Note that imposing F C (dt) = (1 p) 1 F L (dt) = F R (dt), one recovers the equations of Model 1. However, for the applications we have in mind, there is no natural interpretation for such a constraint in Turnbull s model. Moreover, we derive a product-limit estimator for our Model 1. Finally, the asymptotic results below are much simpler and they are obtained under weaker conditions than in Turnbull s model. 3 Asymptotic results In this section the strong uniform convergence and the asymptotic normality for the estimator of the distribution F T in Model 1 are derived. Moreover, we propose a bootstrap procedure that can be used to build confidence intervals for F T. As in the previous sections, the distributions F T and F C need not be continuous. For simpler notation, hereafter, the subscript T is suppressed when there is no possible confusion. We write F (resp. F, Λ and Λ) instead of F T (resp. F T, ΛT and Λ T ). 3.1 Strong uniform convergence Let H nk be the empirical counterparts of the subdistributions H k, k = 0, 1, 2,, that is n H nk ([0, t]) = 1 {Yi t, A i=k}, k = 0, 1, 2. i=1 Clearly, sup t 0 H nk ([0, t]) H k ([0, t]) 0, almost surely. We want to prove the strong uniform convergence of the distribution F, that is F([0, t]) F([0, t]) 0, asn, almost surely, sup t I where I = {t : H 0 ([t, ])+ph 1 ([t, ]) > 0}. For this purpose, first we prove the almost sure convergence of the hazard function.
7 PL estimators with left or right censored data 985 Theorem 1 Assume that p (0, 1] and let t = supi. For any σ I, sup Λ([0, t]) Λ ([0, t]) 0, as n, almost surely. 0 t σ Moreover, if t / I and Λ ([0, t )) <, then Λ ([0, t )) Λ ([0, t )), almost surely. The strong uniform convergence of the distribution F follows without any additional assumption. Theorem 2 Assume that p (0, 1]. Then F([0, t]) F([0, t]) 0, as n, almost surely. sup t I With p = 1 one recovers the strong uniform convergence result for the Kaplan-Meier estimator obtained by [Stute and Wang, 1994], [Gill, 1994]. Our alternative proof is simpler. 3.2 Asymptotic normality Here we study the weak convergence of the process n( F F) where F is the product-limit estimator of Model 1. In this case, Λ does no longer have a martingale structure (in t) as in the case of the Nelson-Aalen estimator, that is when p = 1. However, a continuous time submartingale property for Λ can be obtained. This suffices us to extend the techniques of Gill (1983) and to use them in combination with the functional delta-method in order to establish the weak convergence of n( F F) to a Gaussian process. Here, the weak convergence is denoted by. The space D[a, b] of càdlàg functions defined on [a, b] is endowed with the supremum norm and the ball σ field. Theorem 3 Assume that p (0, 1] and define U(t) = n( F([0, t]) F([0, t])), t 0. Let t = supi. a) Let τ be a point in I. Then, U G in D[0, τ], where G is a Gaussian process. b) If t I, but H 0 (dt) <, (1) {H 0 ([t, ]) + ph 1 ([t, ])} 2 [0,t ) then G can be extended to a Gaussian process on [0, t ] and U G in D[0, t ]. The proof of the weak convergence is postponed to the appendix. Note that when t I, condition (1) is equivalent to F T (dt) F T ([t, ]) > 0 and <. (2) F C ([t, ]) [0,t )
8 986 Patilea and Rolin 3.3 Bootstrapping the product-limit estimator Theorem 3 may be used to obtain confidence intervals and confidence bands for F. However, the law of the process G(t) = G(t)/F ((t, ]) being complicated, one may prefer a bootstrap method in order to avoid handling this process in practical applications. Here, a bootstrap sample is obtained by simple random sample with replacement from the set of observations {(Y i, A i ) : 1 i n}. Let {(Yi, A i ) : 1 i n} denote a bootstrap sample and let Hk be the corresponding subdistributions. Apply equations (4) to (6) to obtain the bootstrap estimator F. The following theorem state that the bootstrap works almost surely for our product-limit estimator on any interval [0, τ] such that H 0 ([τ, ]) + ph 1 ([τ, ]) > 0. This result is a simple corollary of Theorem of [Van der Vaart and Wellner, 1996] and it is based on the uniform Hadamard differentiability of the maps involved in the inversion formula of Model 1. Theorem 4 Let τ I and let G(t) be the limit of U(t)/F ((t, ]) in D[0, τ], as obtained from Theorem 3. Then, the process n{ F ([0, t]) F([0, t])}/ F ((t, ]) converges to G in D[0, τ], almost surely. References [Gill, 1994]R. Gill. Lectures on survival analysis. Lecture Notes in Mathematics (Ecole d été de Probabilités de Saint-Flour XXII 1992), pages , [Gu and Zhang, 1993]M.G. Gu and C.-H. Zhang. Asymptotic properties of selfconsistent estimators based on doubly censored data. Ann. Statist., pages , [Huang, 1999]J. Huang. Asymptotic properties of nonparametric estimation based on partly interval-censored data. Statist. Sinica, pages , [Kim, 2003]J.S. Kim. Maximum likelihood estimation for the proportional hazards models with partly interval-censored data. J. R. Stat. Soc. Ser B, pages , [Samuelsen, 1989]S.O. Samuelsen. Asymptotic theory for non-parametric estimators from doubly censored data. Scand. J. Statist., pages 1 21, [Stute and Wang, 1994]W. Stute and J. Wang. The strong law under random censorship. Ann. Statist., pages , [Turnbull and Weiss, 1978]B.W. Turnbull and L. Weiss. A likelihood ratio statistics for testing goodness of fit with randomly censored data. Biometrics, pages , [Turnbull, 1974]B.W. Turnbull. Nonparametric estimation of a survivorship function with doubly censored data. J. Amer. Statist. Assoc., pages , [Van der Vaart and Wellner, 1996]A.W. Van der Vaart and J.A. Wellner. Weak convergence and empirical processes. Springer Verlag, New-York, 1996.
Product-limit estimators of the survival function for two modified forms of currentstatus
Bernoulli 12(5), 2006, 801 819 Product-limit estimators of the survival function for two modified forms of currentstatus data VALENTIN PATILEA 1 and JEAN-MARIE ROLIN 2 1 CREST-ENSAI, Campus de Ker Lann,
More informationEstimation of the Bivariate and Marginal Distributions with Censored Data
Estimation of the Bivariate and Marginal Distributions with Censored Data Michael Akritas and Ingrid Van Keilegom Penn State University and Eindhoven University of Technology May 22, 2 Abstract Two new
More informationStatistics 262: Intermediate Biostatistics Non-parametric Survival Analysis
Statistics 262: Intermediate Biostatistics Non-parametric Survival Analysis Jonathan Taylor & Kristin Cobb Statistics 262: Intermediate Biostatistics p.1/?? Overview of today s class Kaplan-Meier Curve
More informationAsymptotic Properties of Nonparametric Estimation Based on Partly Interval-Censored Data
Asymptotic Properties of Nonparametric Estimation Based on Partly Interval-Censored Data Jian Huang Department of Statistics and Actuarial Science University of Iowa, Iowa City, IA 52242 Email: jian@stat.uiowa.edu
More informationUniversity of California, Berkeley
University of California, Berkeley U.C. Berkeley Division of Biostatistics Working Paper Series Year 24 Paper 153 A Note on Empirical Likelihood Inference of Residual Life Regression Ying Qing Chen Yichuan
More informationA review of some semiparametric regression models with application to scoring
A review of some semiparametric regression models with application to scoring Jean-Loïc Berthet 1 and Valentin Patilea 2 1 ENSAI Campus de Ker-Lann Rue Blaise Pascal - BP 37203 35172 Bruz cedex, France
More informationGOODNESS-OF-FIT TESTS FOR ARCHIMEDEAN COPULA MODELS
Statistica Sinica 20 (2010), 441-453 GOODNESS-OF-FIT TESTS FOR ARCHIMEDEAN COPULA MODELS Antai Wang Georgetown University Medical Center Abstract: In this paper, we propose two tests for parametric models
More informationApproximate Self Consistency for Middle-Censored Data
Approximate Self Consistency for Middle-Censored Data by S. Rao Jammalamadaka Department of Statistics & Applied Probability, University of California, Santa Barbara, CA 93106, USA. and Srikanth K. Iyer
More informationEmpirical likelihood ratio with arbitrarily censored/truncated data by EM algorithm
Empirical likelihood ratio with arbitrarily censored/truncated data by EM algorithm Mai Zhou 1 University of Kentucky, Lexington, KY 40506 USA Summary. Empirical likelihood ratio method (Thomas and Grunkmier
More informationStatistical Analysis of Competing Risks With Missing Causes of Failure
Proceedings 59th ISI World Statistics Congress, 25-3 August 213, Hong Kong (Session STS9) p.1223 Statistical Analysis of Competing Risks With Missing Causes of Failure Isha Dewan 1,3 and Uttara V. Naik-Nimbalkar
More informationAsymptotic Distributions for the Nelson-Aalen and Kaplan-Meier estimators and for test statistics.
Asymptotic Distributions for the Nelson-Aalen and Kaplan-Meier estimators and for test statistics. Dragi Anevski Mathematical Sciences und University November 25, 21 1 Asymptotic distributions for statistical
More informationand Comparison with NPMLE
NONPARAMETRIC BAYES ESTIMATOR OF SURVIVAL FUNCTIONS FOR DOUBLY/INTERVAL CENSORED DATA and Comparison with NPMLE Mai Zhou Department of Statistics, University of Kentucky, Lexington, KY 40506 USA http://ms.uky.edu/
More informationSTAT Sample Problem: General Asymptotic Results
STAT331 1-Sample Problem: General Asymptotic Results In this unit we will consider the 1-sample problem and prove the consistency and asymptotic normality of the Nelson-Aalen estimator of the cumulative
More information11 Survival Analysis and Empirical Likelihood
11 Survival Analysis and Empirical Likelihood The first paper of empirical likelihood is actually about confidence intervals with the Kaplan-Meier estimator (Thomas and Grunkmeier 1979), i.e. deals with
More informationFULL LIKELIHOOD INFERENCES IN THE COX MODEL
October 20, 2007 FULL LIKELIHOOD INFERENCES IN THE COX MODEL BY JIAN-JIAN REN 1 AND MAI ZHOU 2 University of Central Florida and University of Kentucky Abstract We use the empirical likelihood approach
More informationEmpirical Likelihood in Survival Analysis
Empirical Likelihood in Survival Analysis Gang Li 1, Runze Li 2, and Mai Zhou 3 1 Department of Biostatistics, University of California, Los Angeles, CA 90095 vli@ucla.edu 2 Department of Statistics, The
More informationNonparametric estimation of linear functionals of a multivariate distribution under multivariate censoring with applications.
Chapter 1 Nonparametric estimation of linear functionals of a multivariate distribution under multivariate censoring with applications. 1.1. Introduction The study of dependent durations arises in many
More informationEfficiency of Profile/Partial Likelihood in the Cox Model
Efficiency of Profile/Partial Likelihood in the Cox Model Yuichi Hirose School of Mathematics, Statistics and Operations Research, Victoria University of Wellington, New Zealand Summary. This paper shows
More informationEmpirical Processes & Survival Analysis. The Functional Delta Method
STAT/BMI 741 University of Wisconsin-Madison Empirical Processes & Survival Analysis Lecture 3 The Functional Delta Method Lu Mao lmao@biostat.wisc.edu 3-1 Objectives By the end of this lecture, you will
More informationASYMPTOTIC PROPERTIES OF NONPARAMETRIC ESTIMATION BASED ON PARTLY INTERVAL-CENSORED DATA
Statistica Sinica 9(1999), 501-519 ASYMPTOTIC PROPERTIES OF NONPARAMETRIC ESTIMATION BASED ON PARTLY INTERVAL-CENSORED DATA Jian Huang University of Iowa Abstract: We study asymptotic properties of the
More informationConsistency of bootstrap procedures for the nonparametric assessment of noninferiority with random censorship
Consistency of bootstrap procedures for the nonparametric assessment of noninferiority with random censorship Gudrun Freitag 1 and Axel Mun Institut für Mathematische Stochasti Georg-August-Universität
More informationA COMPARISON OF POISSON AND BINOMIAL EMPIRICAL LIKELIHOOD Mai Zhou and Hui Fang University of Kentucky
A COMPARISON OF POISSON AND BINOMIAL EMPIRICAL LIKELIHOOD Mai Zhou and Hui Fang University of Kentucky Empirical likelihood with right censored data were studied by Thomas and Grunkmier (1975), Li (1995),
More informationLocal Linear Estimation with Censored Data
Introduction Main Result Comparison with Related Works June 6, 2011 Introduction Main Result Comparison with Related Works Outline 1 Introduction Self-Consistent Estimators Local Linear Estimation of Regression
More information1 Glivenko-Cantelli type theorems
STA79 Lecture Spring Semester Glivenko-Cantelli type theorems Given i.i.d. observations X,..., X n with unknown distribution function F (t, consider the empirical (sample CDF ˆF n (t = I [Xi t]. n Then
More informationAn augmented inverse probability weighted survival function estimator
An augmented inverse probability weighted survival function estimator Sundarraman Subramanian & Dipankar Bandyopadhyay Abstract We analyze an augmented inverse probability of non-missingness weighted estimator
More informationNONPARAMETRIC BAYES ESTIMATOR OF SURVIVAL FUNCTIONS FOR DOUBLY/INTERVAL CENSORED DATA
Statistica Sinica 14(2004, 533-546 NONPARAMETRIC BAYES ESTIMATOR OF SURVIVAL FUNCTIONS FOR DOUBLY/INTERVAL CENSORED DATA Mai Zhou University of Kentucky Abstract: The non-parametric Bayes estimator with
More information1. Introduction In many biomedical studies, the random survival time of interest is never observed and is only known to lie before an inspection time
ASYMPTOTIC PROPERTIES OF THE GMLE WITH CASE 2 INTERVAL-CENSORED DATA By Qiqing Yu a;1 Anton Schick a, Linxiong Li b;2 and George Y. C. Wong c;3 a Dept. of Mathematical Sciences, Binghamton University,
More informationOn the generalized maximum likelihood estimator of survival function under Koziol Green model
On the generalized maximum likelihood estimator of survival function under Koziol Green model By: Haimeng Zhang, M. Bhaskara Rao, Rupa C. Mitra Zhang, H., Rao, M.B., and Mitra, R.C. (2006). On the generalized
More informationEMPIRICAL ENVELOPE MLE AND LR TESTS. Mai Zhou University of Kentucky
EMPIRICAL ENVELOPE MLE AND LR TESTS Mai Zhou University of Kentucky Summary We study in this paper some nonparametric inference problems where the nonparametric maximum likelihood estimator (NPMLE) are
More informationExercises. (a) Prove that m(t) =
Exercises 1. Lack of memory. Verify that the exponential distribution has the lack of memory property, that is, if T is exponentially distributed with parameter λ > then so is T t given that T > t for
More informationMultistate Modeling and Applications
Multistate Modeling and Applications Yang Yang Department of Statistics University of Michigan, Ann Arbor IBM Research Graduate Student Workshop: Statistics for a Smarter Planet Yang Yang (UM, Ann Arbor)
More informationSurvival Analysis: Weeks 2-3. Lu Tian and Richard Olshen Stanford University
Survival Analysis: Weeks 2-3 Lu Tian and Richard Olshen Stanford University 2 Kaplan-Meier(KM) Estimator Nonparametric estimation of the survival function S(t) = pr(t > t) The nonparametric estimation
More informationCensoring and Truncation - Highlighting the Differences
Censoring and Truncation - Highlighting the Differences Micha Mandel The Hebrew University of Jerusalem, Jerusalem, Israel, 91905 July 9, 2007 Micha Mandel is a Lecturer, Department of Statistics, The
More informationA dimension reduction approach for conditional Kaplan-Meier estimators
A dimension reduction approach for conditional Kaplan-Meier estimators Weiyu Li and Valentin Patilea January 2015 Abstract In many situations, the quantities of interest in survival analysis are smooth,
More informationLecture 3. Truncation, length-bias and prevalence sampling
Lecture 3. Truncation, length-bias and prevalence sampling 3.1 Prevalent sampling Statistical techniques for truncated data have been integrated into survival analysis in last two decades. Truncation in
More informationThe Hopf argument. Yves Coudene. IRMAR, Université Rennes 1, campus beaulieu, bat Rennes cedex, France
The Hopf argument Yves Coudene IRMAR, Université Rennes, campus beaulieu, bat.23 35042 Rennes cedex, France yves.coudene@univ-rennes.fr slightly updated from the published version in Journal of Modern
More informationSTAT331. Combining Martingales, Stochastic Integrals, and Applications to Logrank Test & Cox s Model
STAT331 Combining Martingales, Stochastic Integrals, and Applications to Logrank Test & Cox s Model Because of Theorem 2.5.1 in Fleming and Harrington, see Unit 11: For counting process martingales with
More informationLecture 5 Models and methods for recurrent event data
Lecture 5 Models and methods for recurrent event data Recurrent and multiple events are commonly encountered in longitudinal studies. In this chapter we consider ordered recurrent and multiple events.
More informationInconsistency of the MLE for the joint distribution. of interval censored survival times. and continuous marks
Inconsistency of the MLE for the joint distribution of interval censored survival times and continuous marks By M.H. Maathuis and J.A. Wellner Department of Statistics, University of Washington, Seattle,
More informationEditorial Manager(tm) for Lifetime Data Analysis Manuscript Draft. Manuscript Number: Title: On An Exponential Bound for the Kaplan - Meier Estimator
Editorial Managertm) for Lifetime Data Analysis Manuscript Draft Manuscript Number: Title: On An Exponential Bound for the Kaplan - Meier Estimator Article Type: SI-honor of N. Breslow invitation only)
More informationQuantile Regression for Residual Life and Empirical Likelihood
Quantile Regression for Residual Life and Empirical Likelihood Mai Zhou email: mai@ms.uky.edu Department of Statistics, University of Kentucky, Lexington, KY 40506-0027, USA Jong-Hyeon Jeong email: jeong@nsabp.pitt.edu
More informationLecture 6 PREDICTING SURVIVAL UNDER THE PH MODEL
Lecture 6 PREDICTING SURVIVAL UNDER THE PH MODEL The Cox PH model: λ(t Z) = λ 0 (t) exp(β Z). How do we estimate the survival probability, S z (t) = S(t Z) = P (T > t Z), for an individual with covariates
More informationPart III Measures of Classification Accuracy for the Prediction of Survival Times
Part III Measures of Classification Accuracy for the Prediction of Survival Times Patrick J Heagerty PhD Department of Biostatistics University of Washington 102 ISCB 2010 Session Three Outline Examples
More informationNonparametric Bayes Estimator of Survival Function for Right-Censoring and Left-Truncation Data
Nonparametric Bayes Estimator of Survival Function for Right-Censoring and Left-Truncation Data Mai Zhou and Julia Luan Department of Statistics University of Kentucky Lexington, KY 40506-0027, U.S.A.
More informationSTAT Section 2.1: Basic Inference. Basic Definitions
STAT 518 --- Section 2.1: Basic Inference Basic Definitions Population: The collection of all the individuals of interest. This collection may be or even. Sample: A collection of elements of the population.
More informationFULL LIKELIHOOD INFERENCES IN THE COX MODEL: AN EMPIRICAL LIKELIHOOD APPROACH
FULL LIKELIHOOD INFERENCES IN THE COX MODEL: AN EMPIRICAL LIKELIHOOD APPROACH Jian-Jian Ren 1 and Mai Zhou 2 University of Central Florida and University of Kentucky Abstract: For the regression parameter
More informationST745: Survival Analysis: Nonparametric methods
ST745: Survival Analysis: Nonparametric methods Eric B. Laber Department of Statistics, North Carolina State University February 5, 2015 The KM estimator is used ubiquitously in medical studies to estimate
More informationSTAT331 Lebesgue-Stieltjes Integrals, Martingales, Counting Processes
STAT331 Lebesgue-Stieltjes Integrals, Martingales, Counting Processes This section introduces Lebesgue-Stieltjes integrals, and defines two important stochastic processes: a martingale process and a counting
More informationSOME CONVERSE LIMIT THEOREMS FOR EXCHANGEABLE BOOTSTRAPS
SOME CONVERSE LIMIT THEOREMS OR EXCHANGEABLE BOOTSTRAPS Jon A. Wellner University of Washington The bootstrap Glivenko-Cantelli and bootstrap Donsker theorems of Giné and Zinn (990) contain both necessary
More informationYou know I m not goin diss you on the internet Cause my mama taught me better than that I m a survivor (What?) I m not goin give up (What?
You know I m not goin diss you on the internet Cause my mama taught me better than that I m a survivor (What?) I m not goin give up (What?) I m not goin stop (What?) I m goin work harder (What?) Sir David
More informationAnalysis of Time-to-Event Data: Chapter 2 - Nonparametric estimation of functions of survival time
Analysis of Time-to-Event Data: Chapter 2 - Nonparametric estimation of functions of survival time Steffen Unkel Department of Medical Statistics University Medical Center Göttingen, Germany Winter term
More informationPairwise dependence diagnostics for clustered failure-time data
Biometrika Advance Access published May 13, 27 Biometrika (27), pp. 1 15 27 Biometrika Trust Printed in Great Britain doi:1.193/biomet/asm24 Pairwise dependence diagnostics for clustered failure-time data
More informationNonparametric Model Construction
Nonparametric Model Construction Chapters 4 and 12 Stat 477 - Loss Models Chapters 4 and 12 (Stat 477) Nonparametric Model Construction Brian Hartman - BYU 1 / 28 Types of data Types of data For non-life
More informationEMPIRICAL LIKELIHOOD AND DIFFERENTIABLE FUNCTIONALS
University of Kentucky UKnowledge Theses and Dissertations--Statistics Statistics 2016 EMPIRICAL LIKELIHOOD AND DIFFERENTIABLE FUNCTIONALS Zhiyuan Shen University of Kentucky, alanshenpku10@gmail.com Digital
More informationEstimating Bivariate Survival Function by Volterra Estimator Using Dynamic Programming Techniques
Journal of Data Science 7(2009), 365-380 Estimating Bivariate Survival Function by Volterra Estimator Using Dynamic Programming Techniques Jiantian Wang and Pablo Zafra Kean University Abstract: For estimating
More informationComparing Distribution Functions via Empirical Likelihood
Georgia State University ScholarWorks @ Georgia State University Mathematics and Statistics Faculty Publications Department of Mathematics and Statistics 25 Comparing Distribution Functions via Empirical
More informationApproximate self consistency for middle-censored data
Journal of Statistical Planning and Inference 124 (2004) 75 86 www.elsevier.com/locate/jspi Approximate self consistency for middle-censored data S. Rao Jammalamadaka a;, Srikanth K. Iyer b a Department
More informationApproximation of Survival Function by Taylor Series for General Partly Interval Censored Data
Malaysian Journal of Mathematical Sciences 11(3): 33 315 (217) MALAYSIAN JOURNAL OF MATHEMATICAL SCIENCES Journal homepage: http://einspem.upm.edu.my/journal Approximation of Survival Function by Taylor
More informationPart III. Hypothesis Testing. III.1. Log-rank Test for Right-censored Failure Time Data
1 Part III. Hypothesis Testing III.1. Log-rank Test for Right-censored Failure Time Data Consider a survival study consisting of n independent subjects from p different populations with survival functions
More informationSmooth nonparametric estimation of a quantile function under right censoring using beta kernels
Smooth nonparametric estimation of a quantile function under right censoring using beta kernels Chanseok Park 1 Department of Mathematical Sciences, Clemson University, Clemson, SC 29634 Short Title: Smooth
More informationMeei Pyng Ng 1 and Ray Watson 1
Aust N Z J Stat 444), 2002, 467 478 DEALING WITH TIES IN FAILURE TIME DATA Meei Pyng Ng 1 and Ray Watson 1 University of Melbourne Summary In dealing with ties in failure time data the mechanism by which
More informationGoodness-of-Fit Tests for Time Series Models: A Score-Marked Empirical Process Approach
Goodness-of-Fit Tests for Time Series Models: A Score-Marked Empirical Process Approach By Shiqing Ling Department of Mathematics Hong Kong University of Science and Technology Let {y t : t = 0, ±1, ±2,
More informationLikelihood ratio confidence bands in nonparametric regression with censored data
Likelihood ratio confidence bands in nonparametric regression with censored data Gang Li University of California at Los Angeles Department of Biostatistics Ingrid Van Keilegom Eindhoven University of
More informationAnalysis of competing risks data and simulation of data following predened subdistribution hazards
Analysis of competing risks data and simulation of data following predened subdistribution hazards Bernhard Haller Institut für Medizinische Statistik und Epidemiologie Technische Universität München 27.05.2013
More informationAFT Models and Empirical Likelihood
AFT Models and Empirical Likelihood Mai Zhou Department of Statistics, University of Kentucky Collaborators: Gang Li (UCLA); A. Bathke; M. Kim (Kentucky) Accelerated Failure Time (AFT) models: Y = log(t
More informationA GENERALIZED ADDITIVE REGRESSION MODEL FOR SURVIVAL TIMES 1. By Thomas H. Scheike University of Copenhagen
The Annals of Statistics 21, Vol. 29, No. 5, 1344 136 A GENERALIZED ADDITIVE REGRESSION MODEL FOR SURVIVAL TIMES 1 By Thomas H. Scheike University of Copenhagen We present a non-parametric survival model
More informationAsymptotic Properties of Kaplan-Meier Estimator. for Censored Dependent Data. Zongwu Cai. Department of Mathematics
To appear in Statist. Probab. Letters, 997 Asymptotic Properties of Kaplan-Meier Estimator for Censored Dependent Data by Zongwu Cai Department of Mathematics Southwest Missouri State University Springeld,
More informationGOODNESS-OF-FIT TEST FOR RANDOMLY CENSORED DATA BASED ON MAXIMUM CORRELATION. Ewa Strzalkowska-Kominiak and Aurea Grané (1)
Working Paper 4-2 Statistics and Econometrics Series (4) July 24 Departamento de Estadística Universidad Carlos III de Madrid Calle Madrid, 26 2893 Getafe (Spain) Fax (34) 9 624-98-49 GOODNESS-OF-FIT TEST
More informationMAS3301 / MAS8311 Biostatistics Part II: Survival
MAS3301 / MAS8311 Biostatistics Part II: Survival M. Farrow School of Mathematics and Statistics Newcastle University Semester 2, 2009-10 1 13 The Cox proportional hazards model 13.1 Introduction In the
More informationStep-Stress Models and Associated Inference
Department of Mathematics & Statistics Indian Institute of Technology Kanpur August 19, 2014 Outline Accelerated Life Test 1 Accelerated Life Test 2 3 4 5 6 7 Outline Accelerated Life Test 1 Accelerated
More informationChapter 2 Inference on Mean Residual Life-Overview
Chapter 2 Inference on Mean Residual Life-Overview Statistical inference based on the remaining lifetimes would be intuitively more appealing than the popular hazard function defined as the risk of immediate
More informationSTAT331. Cox s Proportional Hazards Model
STAT331 Cox s Proportional Hazards Model In this unit we introduce Cox s proportional hazards (Cox s PH) model, give a heuristic development of the partial likelihood function, and discuss adaptations
More informationSize and Shape of Confidence Regions from Extended Empirical Likelihood Tests
Biometrika (2014),,, pp. 1 13 C 2014 Biometrika Trust Printed in Great Britain Size and Shape of Confidence Regions from Extended Empirical Likelihood Tests BY M. ZHOU Department of Statistics, University
More informationMAS3301 / MAS8311 Biostatistics Part II: Survival
MAS330 / MAS83 Biostatistics Part II: Survival M. Farrow School of Mathematics and Statistics Newcastle University Semester 2, 2009-0 8 Parametric models 8. Introduction In the last few sections (the KM
More informationEstimation of Conditional Kendall s Tau for Bivariate Interval Censored Data
Communications for Statistical Applications and Methods 2015, Vol. 22, No. 6, 599 604 DOI: http://dx.doi.org/10.5351/csam.2015.22.6.599 Print ISSN 2287-7843 / Online ISSN 2383-4757 Estimation of Conditional
More informationPreservation Theorems for Glivenko-Cantelli and Uniform Glivenko-Cantelli Classes
Preservation Theorems for Glivenko-Cantelli and Uniform Glivenko-Cantelli Classes This is page 5 Printer: Opaque this Aad van der Vaart and Jon A. Wellner ABSTRACT We show that the P Glivenko property
More informationOn the estimation of the entropy rate of finite Markov chains
On the estimation of the entropy rate of finite Markov chains Gabriela Ciuperca 1 and Valerie Girardin 2 1 Université LYON I, LaPCS, 50 Av. Tony-Garnier, 69366 Lyon cedex 07, France, gabriela.ciuperca@pop.univ-lyon1.fr
More informationProduct-limit estimators of the gap time distribution of a renewal process under different sampling patterns
Product-limit estimators of the gap time distribution of a renewal process under different sampling patterns arxiv:13.182v1 [stat.ap] 28 Feb 21 Richard D. Gill Department of Mathematics University of Leiden
More informationChapter 7: Hypothesis testing
Chapter 7: Hypothesis testing Hypothesis testing is typically done based on the cumulative hazard function. Here we ll use the Nelson-Aalen estimate of the cumulative hazard. The survival function is used
More informationStatistical Methods for Alzheimer s Disease Studies
Statistical Methods for Alzheimer s Disease Studies Rebecca A. Betensky, Ph.D. Department of Biostatistics, Harvard T.H. Chan School of Public Health July 19, 2016 1/37 OUTLINE 1 Statistical collaborations
More informationAsymptotic statistics using the Functional Delta Method
Quantiles, Order Statistics and L-Statsitics TU Kaiserslautern 15. Februar 2015 Motivation Functional The delta method introduced in chapter 3 is an useful technique to turn the weak convergence of random
More informationLecture 2: Martingale theory for univariate survival analysis
Lecture 2: Martingale theory for univariate survival analysis In this lecture T is assumed to be a continuous failure time. A core question in this lecture is how to develop asymptotic properties when
More informationPENALIZED LIKELIHOOD PARAMETER ESTIMATION FOR ADDITIVE HAZARD MODELS WITH INTERVAL CENSORED DATA
PENALIZED LIKELIHOOD PARAMETER ESTIMATION FOR ADDITIVE HAZARD MODELS WITH INTERVAL CENSORED DATA Kasun Rathnayake ; A/Prof Jun Ma Department of Statistics Faculty of Science and Engineering Macquarie University
More informationUNIVERSITY OF CALIFORNIA, SAN DIEGO
UNIVERSITY OF CALIFORNIA, SAN DIEGO Estimation of the primary hazard ratio in the presence of a secondary covariate with non-proportional hazards An undergraduate honors thesis submitted to the Department
More informationEMPIRICAL LIKELIHOOD ANALYSIS FOR THE HETEROSCEDASTIC ACCELERATED FAILURE TIME MODEL
Statistica Sinica 22 (2012), 295-316 doi:http://dx.doi.org/10.5705/ss.2010.190 EMPIRICAL LIKELIHOOD ANALYSIS FOR THE HETEROSCEDASTIC ACCELERATED FAILURE TIME MODEL Mai Zhou 1, Mi-Ok Kim 2, and Arne C.
More informationApplication of Time-to-Event Methods in the Assessment of Safety in Clinical Trials
Application of Time-to-Event Methods in the Assessment of Safety in Clinical Trials Progress, Updates, Problems William Jen Hoe Koh May 9, 2013 Overview Marginal vs Conditional What is TMLE? Key Estimation
More informationA Recursive Formula for the Kaplan-Meier Estimator with Mean Constraints
Noname manuscript No. (will be inserted by the editor) A Recursive Formula for the Kaplan-Meier Estimator with Mean Constraints Mai Zhou Yifan Yang Received: date / Accepted: date Abstract In this note
More informationOrder restricted inference for comparing the cumulative incidence of a competing risk over several populations
IMS Collections Beyond Parametrics in Interdisciplinary Research: Festschrift in Honor of Professor Pranab K. Sen Vol. 1 (2008) 50 61 c Institute of Mathematical Statistics, 2008 DOI: 10.1214/193940307000000040
More informationOn consistency of Kendall s tau under censoring
Biometria (28), 95, 4,pp. 997 11 C 28 Biometria Trust Printed in Great Britain doi: 1.193/biomet/asn37 Advance Access publication 17 September 28 On consistency of Kendall s tau under censoring BY DAVID
More informationCOMPUTE CENSORED EMPIRICAL LIKELIHOOD RATIO BY SEQUENTIAL QUADRATIC PROGRAMMING Kun Chen and Mai Zhou University of Kentucky
COMPUTE CENSORED EMPIRICAL LIKELIHOOD RATIO BY SEQUENTIAL QUADRATIC PROGRAMMING Kun Chen and Mai Zhou University of Kentucky Summary Empirical likelihood ratio method (Thomas and Grunkmier 975, Owen 988,
More informationUnderstanding product integration. A talk about teaching survival analysis.
Understanding product integration. A talk about teaching survival analysis. Jan Beyersmann, Arthur Allignol, Martin Schumacher. Freiburg, Germany DFG Research Unit FOR 534 jan@fdm.uni-freiburg.de It is
More informationChapter 3. Chord Length Estimation. 3.1 Introduction
Chapter 3 Chord Length Estimation 3.1 Introduction Consider a random closed set W R 2 which we observe through a bounded window B. Important characteristics of the probability distribution of a random
More informationOn the Breslow estimator
Lifetime Data Anal (27) 13:471 48 DOI 1.17/s1985-7-948-y On the Breslow estimator D. Y. Lin Received: 5 April 27 / Accepted: 16 July 27 / Published online: 2 September 27 Springer Science+Business Media,
More informationChapter 4 Fall Notations: t 1 < t 2 < < t D, D unique death times. d j = # deaths at t j = n. Y j = # at risk /alive at t j = n
Bios 323: Applied Survival Analysis Qingxia (Cindy) Chen Chapter 4 Fall 2012 4.2 Estimators of the survival and cumulative hazard functions for RC data Suppose X is a continuous random failure time with
More informationEfficient Semiparametric Estimators via Modified Profile Likelihood in Frailty & Accelerated-Failure Models
NIH Talk, September 03 Efficient Semiparametric Estimators via Modified Profile Likelihood in Frailty & Accelerated-Failure Models Eric Slud, Math Dept, Univ of Maryland Ongoing joint project with Ilia
More informationLectures on Survival Analysis
1 Lectures on Survival Analysis Richard D. Gill Mathematical Institute, University Utrecht, Budapestlaan 6, 3584 CD Utrecht, Netherlands. gill@math.ruu.nl To appear in: Ecole d Eté de Probabilités de Saint
More informationConcentration behavior of the penalized least squares estimator
Concentration behavior of the penalized least squares estimator Penalized least squares behavior arxiv:1511.08698v2 [math.st] 19 Oct 2016 Alan Muro and Sara van de Geer {muro,geer}@stat.math.ethz.ch Seminar
More informationEstimation for Modified Data
Definition. Estimation for Modified Data 1. Empirical distribution for complete individual data (section 11.) An observation X is truncated from below ( left truncated) at d if when it is at or below d
More informationLecture 22 Survival Analysis: An Introduction
University of Illinois Department of Economics Spring 2017 Econ 574 Roger Koenker Lecture 22 Survival Analysis: An Introduction There is considerable interest among economists in models of durations, which
More informationTime Series Analysis. Asymptotic Results for Spatial ARMA Models
Communications in Statistics Theory Methods, 35: 67 688, 2006 Copyright Taylor & Francis Group, LLC ISSN: 036-0926 print/532-45x online DOI: 0.080/036092050049893 Time Series Analysis Asymptotic Results
More information